[Bug 286502] RTLD deadlocks during IFUNC resolution

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=286502

--- Comment #4 from Dapeng Gao <dapeng.gao@cl.cam.ac.uk> ---
(In reply to Konstantin Belousov from comment #3)

I'm a little confused.

If the first symbol in foo.so to be resolved by _rtld_bind were just a regular
STT_FUNC, filter loading would happen there and no deadlock would occur when
the ifunc is resolved in the future.

So, the resolver does not inherently cause the filter loading, and deadlock may
or may not happen depending on whether the resolver is the first resolved
symbol. This seems rather undesirable.

-- 
You are receiving this mail because:
You are the assignee for the bug.